Ripping embedded subtitles

This time I am thinking of giving up capturing subs already embedded on an old film I have.

This time I went through all the paces, to help Subrip scan subtitles correctly. I even zoomed on the subtitle area, using Procoder, feeding just the subtitles so letters could be larger.

But it seems to be useless. The program is too dumb, even dumber than the first paper scanners were.

Even if you get close to the tone of subtitle letters, when you run it it fails to recognize it. So you end up writing down all the letters. The program does not seem to learn how to OCR!

I wonder what else can be done to get things through. Any suggestions?